AI giant Anthropic has now filed sworn declarations in a California federal court disputing the Pentagon’s claim that it poses an “unacceptable risk to national security”. According to a report by Techcrunch, the filings submitted along with Anthropic’s reply brief argue that the government’s case rests on technical misunderstandings and mischaracterisations of the company’s stance during negotiations. A hearing is now scheduled for March 24, before Judge Rita Lin in San Francisco.
Anthropic vs Pentagon: Dispute over military use of AI
The conflict between Anthropic and Pentagon started in late February when US President Donald Trump and Defense secretary Pete Hegseth announced that they were cutting ties with the AI giant after Anthropic refused to allow unrestricted military use of its AI technology. The Pentagon subsequently designated Anthropic as a supply-chain risk, the first time such a designation has been applied to an American AI company.
Anthropic’s policy chief challenges Pentagon’s claims
As per the TechCrunch report, in her declaration, Sarah Heck, Anthropic’s head of policy and a former White House official, rejected the government’s assertion that the company demanded approval over military operations. “At no time during Anthropic’s negotiations with the Department did I or any other Anthropic employee state that the company wanted that kind of role,” she wrote. Heck also noted that concerns about Anthropic disabling its technology mid-operation were never raised during talks, appearing only in the Pentagon’s court filings.She highlighted a March 4 email from Pentagon Under Secretary Emil Michael to Anthropic CEO Dario Amodei, in which Michael said the two sides were “very close” on the very issues—autonomous weapons and mass surveillance—that the government later cited as evidence of a security threat.Along with this, Anthropic’s head of public sector Thiyagu Ramasamy, also filed a declaration, disputing claims that the company could interfere with military operations. He explained that once Anthropic’s Claude AI is deployed inside government-secured, “air-gapped” systems, the company has no access, no kill switch, and no backdoor. Any updates require Pentagon approval and installation. Ramasamy further argued that Anthropic’s cleared personnel—vetted through U.S. government security processes—make it unique among AI firms working in classified environments.Anthropic’s lawsuit contends that the Pentagon’s designation amounts to retaliation for the company’s public stance on AI safety, violating its First Amendment rights. The government, in a 40-page filing earlier this week, rejected that framing, insisting the designation was a straightforward national security decision and not punishment for Anthropic’s views.
